ABSTRACT
A fundamental component of a life cycle analysis of a bridge is the modelling of asset condition post-construction. To enhance the accuracy of bridge deterioration models, several studies have calibrated models based on cohorts that incorporate structural characteristics and local factors. However, it is common that these characteristics are independently modelled to avoid sparse cohorts. In this study, a bridge element deterioration model is developed that can simultaneously incorporate multiple structural and local characteristics. The incorporation of multiple co-variates is made possible by an approach that exploits the multiple defect deterioration model proposed by Calvert et al. (2021). The model is tested on a cohort of 36,075 metallic bridge girders from bridges on the British railway network. The modelling approach is shown to provide statistically significant improvements in the prediction accuracy of deterioration. Moreover, a life cycle analysis outlines the significant cost and condition differences between benign and aggressive cohorts.
